1

How did we get here from there? The transformation of analytic philosophy1
Timothy Williamson

[Forthcoming in Belgrade Philosophical Annual]

Opponents of analytic philosophy often associate it with logical positivism. From a historical point of
view, it is clear that one main strand in the development of the broad tradition known as analytic
philosophy was indeed the logical positivism of the Vienna Circle, with its austerely verificationist
principle of significance and its exclusion of metaphysics as cognitively meaningless. Another main
strand in the development of the analytic tradition, ordinary language philosophy, tended to be
almost equally suspicious of the ways in which metaphysicians made free with ordinary words, far
from the everyday contexts of use on which their meaning was supposed to depend. Despite that
history, however, recent decades have seen the growth and flourishing of boldly speculative
metaphysics within the analytic tradition. Far from being inhibited by logical positivist or ordinary
language scruples, such analytic metaphysics might be described by those unsympathetic to it as
pre-critical, ranging far outside the domain of our experience, closer in spirit to Leibniz than to Kant.
How did a species of philosophy with so much anti-metaphysics in its gene pool evolve so
quickly to the opposite extreme? Enough time has passed for us now to start achieving the historical
perspective necessary to answer the question in a systematic way. That project is too extensive to be
properly carried out in less than a book. In this paper, I attempt no more than to make some
informal and unsystematic remarks on the transformation of analytic philosophy. Especially in
section II, I write as someone who lived through the latter stages of the process, and concentrate on

2
the parts of which I have the closest knowledge. That will at least provide some sense of what it was
like to experience the transformation at the time, from a broadly sympathetic point of view. I close
with a few sketchy remarks on the historiography of recent analytic philosophy, in section III.

The central, most influential figure in the development of analytic metaphysics over the final quarter
of the twentieth century, and the contemporary philosopher most cited within recent analytic
philosophy, is undoubtedly David Lewis, also known as the machine in the ghost for his eerie
computational power, mechanical diction, faint air of detachment from ordinary life, and beard from
another era (by contrast with the ghost in the machine, Gilbert Ryles summary description of the
immaterial Cartesian ego in the clockwork Cartesian body). The prize specimen of Lewiss speculative
metaphysics is in turn his notorious doctrine of modal realism, according to which there are infinitely
many possible worlds, mutually disconnected spatiotemporal systems each as real and concrete as
our own actual world (Lewis 1986). For Lewis, strictly and literally there are talking donkeys, because
there could have been talking donkeys (as we may all agree), so some possible world contains talking
donkeys, which are just as real, alive, and made of flesh and blood as any donkey you have ever
seen. Of course, those other worlds are not open to our observation; there are no trans-world
telescopes. Lewis postulates them because they follow from his modal realism, which he regards as
the best theory of possibility, necessity, and related phenomena, in respect of simplicity, strength,
elegance, and explanatory power: to use C.S. Peirces term broadly, Lewiss argument for modal
realism is abductive. In a way, Lewis takes non-actual possible worlds even more seriously than did
Leibniz, for whom they are merely unrealized ideas in the mind of God. Leibnizs God realized only

3
the best of all possible worlds, whereas all of Lewiss possible worlds are equally realized. We can
take Lewiss modal realism as a case study for the resurgence of speculative metaphysics in
contemporary analytic philosophy.
Some philosophers treat any appeal to possible worlds at all as a metaphysical extravagance.
But that is a mistake, for some theories treat possible worlds as merely abstract objects or
representations, harmlessly built from this-worldly materials. Indeed, Rudolf Carnap, the logical
positivist anti-metaphysician par excellence, explicitly compared to Leibnizs possible worlds his
state-descriptions, maximal consistent classes of sentences used in his semantics for languages with
modal operators such as possibly and necessarily. He also compared them with the possible states
of affairs in Wittgensteins Tractatus Logico-Philosophicus (Carnap 1947, p. 9). For Carnap, necessity
is a purely intra-linguistic matter of truth guaranteed by meaning, and possibility is a
correspondingly semantic form of consistency. Lewiss modal realism has always been a minority
view, even amongst those analytic metaphysicians who work with possible worlds of some sort. It is
an example of extreme metaphysics.
Where did Lewiss modal realism come from? It already appears in one of his earliest
published papers (Lewis 1968). Exciting developments in modal logic, culminating in the work of Saul
Kripke, had already made the idea of possible worlds central to the understanding of languages with
modal operators (Kripke 1963). Necessarily is understood as in every possible world and possibly
as in some possible world. For technical mathematical reasons, it turned out to be more fruitful not
to equate possible worlds with Carnaps state-descriptions, or other such representational entities,
but instead to leave their nature unconstrained when characterizing models for the modal language
(Williamson 2013, pp. 81-4). That did not enforce a more metaphysically speculative conception of
possible worlds, but it made space for one.
Another significant factor was the development of tense logic, above all by Arthur Prior. He
was acutely aware of the strong structural analogies between modal logic and tense logic, with

4
possible worlds playing the same role in the modal case as moments of time play in the temporal
case (Prior 1957). An orthographically identical operator could be read as necessarily in modal logic
and as always (or always in the past or always in the future) in tense logic. When one read
formulas of the logic in temporal terms, they expressed blatantly metaphysical principles about the
structure of time and of existence in time. Those readings provided templates for more metaphysical
readings of the same formulas in modal terms, on which they expressed analogous metaphysical
principles about the structure of possibility and of possible or necessary existence. Carnaps intralinguistic modalities could then be replaced by Kripkes metaphysical modalities, which concern how
things really could have been.
But developments within modal logic alone cannot fully explain Lewiss modal realism. For
Kripke and other leading modal logicians did not go down the modal realist road. Indeed, Lewiss
modal realist semantics for modal languages introduces messy complications that from a purely
technical point of view are quite unmotivated, even by Priors analogy between time and modality.
In particular, Lewis postulates that no individual exists in more than one world, the analogue of the
highly implausible postulate that no individual exists for more than one moment. In order to make
sense of the common sense idea that one could have acted differently, Lewis then has to introduce
an elaborate theory of counterparthood relations between distinct but similar individuals in
different possible worlds. Those complications were motivated by Lewiss prior commitments.
At this point, one must note that Lewiss doctoral supervisor at Harvard in the mid-1960s
was Willard van Orman Quine, the leading critic of modal logic. Quine was especially critical of
quantified modal logic, as developed by Carnap and Ruth Barcan Marcus, which allows one in effect
to reason about the properties and relations that particular individuals could have, as opposed to
just the general states of affairs that could obtain. Originally, Quine had tried to prove that
quantified modal logic is technically flawed to the point of incoherence, that it collapses the modal
distinctions between possibility, necessity, and actuality. As it became clear that his purely formal

5
arguments were technically flawed, Quine gradually switched his line of attack to the informal
intelligibility of quantified modal formulas. His standard of intelligibility in logic was austere: firstorder non-modal logic, roughly, that of the logical constants not, and, or, everything,
something, and is. For Quine, logic is first-order non-modal logic. Lewis assumed modal realism
because it permits the reduction by translation of a quantified modal language to a first-order nonmodal language in which one talks about worlds and individuals in those worlds. Crucially, Lewiss
modal realism gave him a way of informally explaining what a possible world is in non-modal terms:
roughly, a spatiotemporal system; the individuals in such a system are spatiotemporally connected
to each other and to nothing outside the system. Lewis thereby aimed to make quantified modal
logic intelligible by his teachers standards. Since much ordinary discourse in natural language
involves expressive resources at least as great as those of quantified modal logic can and must
are common words, to say the least Lewiss procedure is also motivated by a principle of charity
(Lewis 1974), which Quine explicitly endorsed: prefer an interpretation of a natural language on
which speakers are being sensible to one on which they are being silly (Quine 1960, p. 59). Possible
worlds other than our own and their inhabitants are needed to make true ordinary statements
about what could have been but isnt. Lewis gave an ingenious solution to the Quinean problem of
charitably interpreting quantified modal discourse, on its own Quinean terms, even though Quine
rejected the gift.
Quines lack of interest in modal realism was no anti-metaphysical stance. He did as much as
anyone to put ontology as a branch of metaphysics on the map of analytic philosophy, and his
conception of philosophy as continuous with natural science overtly involved a naturalistic approach
to metaphysical theorizing. One might rather be tempted to suggest that he rejected modal realism
because it lacked support from natural science. However, Quine did not require each point of a
metaphysical theory to receive its own specific support from natural science. For instance, although
he took mathematics with its ontological commitment to abstract objects such as sets or numbers
to be holistically justified by its applications in natural science, he was well aware that the power

6
of the standard axioms of set theory goes far beyond the needs of natural science, but still regarded
them as a legitimate rounding out of the fragment actually used in scientific applications, justified by
its simplicity, elegance, and other such virtues. Thus Quines justification of mathematics is
abductive, in a similar spirit to Lewiss justification of modal realism. No doubt, if Quine had felt
some tension between modal realism and current natural scientific theory, he would have treated
that as a good reason to reject modal realism, but so might Lewis himself. However, neither of them
seems to have felt such a tension. In my view, there is in fact such a tension, or inconsistency, but
the argument for it must be made with some delicacy and was not generally recognized at the time
(Williamson 2013, pp. xii, 17 and 2014, pp. 744-6). Perhaps Quine simply gave modal realism the
same incredulous stare that so many other analytic metaphysicians have given it (Lewis 1986: 1335). Moreover, Quine encountered modal realism at a stage of his career when he was not disposed
to accept radical revisions of his views from outside; publicly retracting his well-entrenched
signature scepticism about quantified modal logic would have been a bitter pill to swallow. He
showed a similar lack of interest even in the purely technical development of the model theory of
modal logic, and in particular of quantified modal logic, by Kripke and others, which is a piece of
regular mathematics, no more vulnerable to Quines concerns about intelligibility than any other
piece of mathematics.
The example of Quine is a salutary reminder that the analytic tradition, as normally
understood, has never been a metaphysics-free zone. Before Quine, Bertrand Russell was a major
figure in the analytic tradition blatantly engaged in metaphysical theorizing. Indeed, F.H. Bradley
may well be right that critiques of metaphysics themselves depend on contentious metaphysical
assumptions (Bradley 1893, pp. 1-2). Nevertheless, the role and status of metaphysics have changed
in significant ways over the history of the analytic tradition, and our concern is with those ways.
Lewiss case for his modal realism itself evolved over time. In the original 1968 paper, the
emphasis is on the relation between modal and non-modal languages, and the clarity to be achieved

7
in modal logic by translating the modal language into the non-modal language of Lewiss counterpart
theory (the precursor of his modal realism). His postulates for counterpart theory are used to
validate elementary principles of modal logic, but they also clarify the metaphysical picture. For
instance, the postulate that nothing is in two worlds has the advantage, according to Lewis, that it
answers the question of the identity of individuals across possible worlds at a stroke, with a uniform
negative. That is his answer to Quines complaints about the obscurity of trans-world individuation
(he cites Quine 1960, p. 245). By the time he wrote what became the canonical case for modal
realism, his book On the Plurality of Worlds (Lewis 1986), based largely on his 1984 John Locke
lectures at Oxford, Lewiss perspective had changed. He talks much less about linguistic matters, and
much more about the abductive advantages of modal realism as a theoretical framework for
explaining a variety of phenomena, many of them non-linguistic. Faced with some objections to
specific translation schemes between the language of quantified modal logic (in effect, a
formalization of ordinary modal language) and the language of counterpart theory, he tells us not to
worry about the details, but instead to abandon the language of quantified modal logic and do our
metaphysical theorizing directly in the more perspicuous language of counterpart theory (Lewis
1986, pp. 12-13). Clearly, from 1968 to 1986 the balance of power in Lewiss work swung towards
metaphysics and away from the philosophy of language.
Writing in 1981, Lewis described a reasonable goal for a philosopher as bringing ones
opinions into stable equilibrium. The trouble with losing our moorings in everyday common sense,
according to him, is not that common sense is infallible but that we do not achieve stable
equilibrium, since we keep reverting to something like our everyday opinions (Lewis 1983a, p. x). He
requires the equilibrium to be stable under theoretical reflection. In principle, this is not radically
different from a Quinean methodology for philosophy and science together of adjustments to ease
tensions in ones web of beliefs (Quine 1951). For Lewis, modal realism does better than other
theories of modal metaphysics with respect to stability under reflection, given our other beliefs. Of
course, many philosophers would insist that to adopt modal realism is to lose ones moorings in

8
everyday common sense, but Lewis was adept at interpreting the putative deliverances of everyday
common sense in ways that made them consistent with modal realism, often by postulating large
measures of tacit contextual restriction in the utterances that gave them voice. Although he
admitted some disagreements between modal realism and common sense, for instance on whether
(speaking unrestrictedly) there are talking donkeys, he managed to steer the disagreements into
abstruse areas that might plausibly be regarded as of low priority for common sense. Whereas critics
condemned Lewis for his extravagant departures from common sense, he saw his modal realism as
part of his defence of common sense just the way Berkeley saw his subjective idealism.
In practice, the process of bringing ones opinions into stable equilibrium will involve
extensive reflection on what ones opinions actually are. Since ones general beliefs are typically
presented to one as expressed by sentences, whose underlying semantic structure is not perfectly
transparent, in reflecting on what it is one believes one is drawn into semantic reflection on ones
own language or, as some would have it, reflection on ones own conceptual system.
A natural comparison is between Lewiss Quinean or at least post-Quinean methodology and
the methodology of Peter Strawson, Quines leading opponent from the tradition of ordinary
language philosophy. By the late 1950s, however, ordinary language philosophy was no longer an
apt phrase for what Strawson was doing. He was working in a far more abstract and systematic way
than that phrase suggests. His concern was with very general structural features of ordinary thought
and talk, such as the distinction between subject and predicate, rather than with the fine detail of
ordinary usage. In 1959 he published Individuals, subtitled An Essay in Descriptive Metaphysics, a
major monograph widely felt at the time to mark a turning-point in the rehabilitation of metaphysics
within analytic philosophy. Strawson contrasted descriptive metaphysics with the wilder revisionary
metaphysics, which despite its partial vision is nevertheless useful when at the service of
descriptive metaphysics (Strawson 1959, p. 9). Revisionary metaphysics can help at the periphery of
our thinking, but goes astray when it tries to revise a massive central core of human thinking which

9
has no history because there are categories and concepts which, in their most fundamental
character, change not at all (ibid., p. 10). What descriptive metaphysics is supposed to describe are
the conceptual connexions that constitute the structure of that central core. In Strawsons view, that
structure is not hierarchical, as envisaged by programmes of conceptual analysis with their noncircular definitions by necessary and sufficient conditions. Rather, the descriptive metaphysician
traces conceptual interconnexions on the same level, going round in complex closed curves: the
exploration is horizontal rather than vertical.
How different in kind is Strawsonian metaphysics, which may revise the margins but must
only describe the core of ordinary thinking, from Lewisian metaphysics, which must not lose its
moorings in everyday common sense in its attempt to steer ones opinions into equilibrium? At this
point, it may be worth recalling that Lewis attended lectures by Strawson, amongst others, when he
spent the academic year 1959-60 as a visiting student from Swarthmore College at Oxford, tutored
by Iris Murdoch, a year that resulted in his momentous decision to major in Philosophy rather than,
as he had previously intended, Chemistry. Of course, Strawsons characterization of descriptive
metaphysics as the tracing of conceptual connexions relies on some form of the analytic-synthetic
distinction, which he had defended with his old teacher Paul Grice against Quines massively
influential critique (Quine 1951, Grice and Strawson 1956). Indeed, the complex closed curve of
definitions that Quine traced from analytic round to other semantic terms and back again in his
attempt to show that none of them could be satisfactorily explained was just the sort of explanation
the descriptive metaphysician sought. But on this issue Lewis sided with his earlier teacher against
his later one. Lewiss first book concluded with an attempt to rehabilitate analyticity as truth in all
possible worlds (Lewis 1969, p. 208), to Quines regret in the books Foreword (Quine 1969, p. xii).
Although both Strawson and Lewis accepted an analytic-synthetic distinction, in their philosophical
practice neither of them had much tendency to use it as the sort of glib conversation-stopper it so
often becomes in less resourceful hands. Admittedly, Strawson was more prone than Lewis to
characterize philosophical questions as questions about words or concepts, or as questions about

10
how we must think about a subject matter, rather than about the subject matter itself. But Lewis
himself frequently went metalinguistic, as we have already seen, and Strawson was quite willing to
speak in a ground-level metaphysical idiom, as with remarks such as A person is not an embodied
ego, but an ego might be a disembodied person (Strawson 1959, p. 103). They both moved easily
between object-language and meta-language, as it were, depending on the argumentative needs of
the moment.
Quine insisted on continuity between philosophy and natural science in a way that Strawson
did not. In this respect, Lewis was closer to Quine than to Strawson. In practice, however, natural
science played only a very minor role in the metaphysics of all three philosophers. A good test is
their treatment of the dispute between the three-dimensionalist Aristotelian metaphysics of
enduring continuants and the four-dimensionalist metaphysics of occurrents composed of
successive time-slices, often associated with Einsteins theory of special relativity. Predictably,
Strawson is a three-dimensionalist while Quine and Lewis are four-dimensionalists. But what is
striking is how little Quine and Lewis made of special relativity in their cases for fourdimensionalism. In Word and Object, Quine emphasized the advantages in logical smoothness of
treating space and time on a par (Quine 1960, pp. 170-2). He adds only as a convenient afterthought
that Einsteins discovery of special relativity leaves no reasonable alternative to treating time as
spacelike, but then immediately points out that the logical benefits of doing so are independent of
Einsteins principle and adds, with references, the idea of paraphrasing tensed sentences into terms
of eternal relations of things to times was clear enough before Einstein (ibid., p. 172). In On the
Plurality of Worlds, Lewiss case for four-dimensionalism does not mention Einstein or depend on
modern science at all; it just relies on rather shaky old-fashioned purely metaphysical reasoning
about temporary intrinsic properties, such as shapes (Lewis 1986, pp. 202-5). Presumably, Lewis was
not convinced that special relativity really did leave no reasonable alternative to treating time as
spacelike, otherwise he would have mentioned such impressive support for his conclusion. Since the
specifics of Lewiss argumentation fail to take seriously an approach that treats temporal operators

11
as explanatorily basic, it is hard to avoid the impression that what was really decisive with him was a
somewhat Quinean preconception about the proper sort of language for doing metaphysics in,
something close to the language of mathematics. Of course, glancing through the pages of On the
Plurality of Worlds, one sees them filled with formula-free English prose, just like the pages of
Individuals. Nevertheless, the systematicity at which Lewis aims is modelled on that of a scientific
theory articulated in a mathematical language. The systematicity at which Strawson aims is
different; it is that of a satisfying general account in English itself, or some other natural language.
The sort of formal logical smoothness that Quine and Lewis valued so highly, Strawson regarded as a
trap. In this less standard sense, despite the similarities, and even in their metaphysics, Strawson
remained at heart an ordinary language philosopher, and Lewis at heart an ideal language
philosopher. Each of them was reluctant to disagree extensively with common sense (or natural
science, for that matter), but left some room for manoeuvre by acknowledging a belt of revisable
opinions on the periphery of common sense. As a result of quite a subtle difference in intellectual
values, they ended up with radically different theories.
What if Strawson is right, in that Lewiss views conflict with the unchanging core of ordinary
thought? Then those views fail by Lewiss own criterion, since no totality of his opinions that
includes them will be stable under theoretical reflection, for it will also include the unchanging core
of ordinary thought. We cannot expect to keep questions about the general methodological position
in a philosophical debate clinically isolated from questions about who is right and who is wrong on
the specific matters at issue within the debate (Williamson 2007, pp. 210-14).
We can refine our sense of the intellectual options by comparing both Strawson and Lewis
with Kripke. In the 1970s, Kripke and Lewis were often paired as leaders of the possible worlds
revolution. Kripkes essentialism and his defence of quantified modal logic were radical in relation
to Quinean orthodoxy. It took time for them to be understood as articulations of quite ordinary ways
of thinking, so not radical in relation to common sense. Of course, the connection between

12
essentialism and Aristotle already noted by Quine was a clue, since Aristotle always had a strong
claim to be the founder of common sense philosophy. Strawson names Aristotle and Kant as the
great descriptive metaphysicians (1959, p. 9). Kripkes arguments in Naming and Necessity tend to
be based on common sense examples, and he explicitly rejects Lewiss modal realism, in favour of a
more deflationary conception of possible worlds as possible states of affairs (Kripke 1972). His
metaphysics is arguably not revisionary in Strawsons sense. This feature of it may have been
obscured by his technical achievements in quantified modal logic, based on the mathematical
apparatus of possible worlds semantics, even though it is in itself no more metaphysically
problematic than any other piece of mathematics (Williamson 2013, pp. 81-4).
Although Kripkes metaphysics initially could seem more revisionary than it really was, his
methodology could initially seem more linguistic than it really was. His titles yoked together
semantic and metaphysical terms: Naming and Necessity (1972) and Reference and Existence (2013,
but the book of his John Locke Lectures given at Oxford in 1973 under the same title). Of course, it
was Kripke who played the central role in distinguishing metaphysical from epistemic or semantic
modalities, through his famous examples of contingent truths knowable a priori and of necessary
truths knowable only a posteriori, which made trouble for the then-popular Humean slogan All
necessity is verbal necessity. Even so, there was a diffuse but influential impression in the 1970s
that Kripke had somehow managed to derive apparently substantial metaphysical conclusions about
the specific essential properties of individuals and kinds, from his semantic analysis of modal
language, in particular his insight that names are rigid designators (their designation remains
constant while different possible worlds are considered). Nathan Salmon published a detailed
monograph Reference and Essence (1982) to refute that impression. The title uses and to separate
the semantic term from the metaphysical one, rather than to join them together. The front cover
showed a rabbit being pulled from a hat. Salmon demonstrated that Kripke had relied on
metaphysical premises to derive his metaphysical conclusion (Kripke himself had not claimed
otherwise). That did not make Kripkes argument merely question-begging, for the plausibility of the

13
premises could still be more immediate than the plausibility of the conclusion. Salmons point was
widely accepted. That contributed to an increasingly popular conception of metaphysics as separate
from the philosophy of language.
Just as it was a mistake to regard Kripkes metaphysics as derived from his semantics, it
would be a mistake to regard them as simply orthogonal to each other. For misconceptions in
semantics often induce misconceptions in metaphysics, by causing fallacious metaphysical
arguments to be treated as valid, so that coherent metaphysical views are incorrectly dismissed as
confused or inconsistent. Quines early critique of quantified modal logic is an example. It was crucial
to get straight about the semantics of modal language in order to see ones way through to
defending metaphysical theses of essentialism. Thus Kripkes semantic theory of rigid designators
was after all relevant to his essentialist metaphysics, but its role was negative, in driving off
arguments against essentialism, not positive, in driving forward arguments for essentialism.
There is a more general moral here about the famous linguistic turn, a phrase which has
looked less and less appropriate as a description of mainstream analytic philosophy over recent
decades. Nevertheless, although analytic philosophers are ceasing to regard their central questions
as linguistic or even conceptual in any sense that would distinguish them from questions asked in
other disciplines, the traces of the linguistic turn are not simply being erased. For it has left a rich
legacy of methodological sophistication. In testing the soundness of arguments about non-linguistic
matters, analytic philosophers regularly draw on work in both semantics and pragmatics (Williamson
2007, pp. 46-7). Kripkes work on quantified modal logic is a good example with respect to
semantics.
With respect to pragmatics, the prime exhibit is the work of Strawsons teacher, Paul Grice,
on conversational implicature (1961, 1975). If you comment after my lecture Williamson was sober
this afternoon, you imply that I am often drunk in the afternoon, even though that is not a
precondition for the truth of what you said: it is even true if I am a scrupulous teetotaller. Grice

14
developed a powerful theoretical apparatus for analysing such effects. Although this work emerged
from within the Oxford of ordinary language philosophy, it made an important contribution to
undermining the methodology of such philosophy. For ordinary language philosophy involved a
focus on what we would say in various conversational contexts. By analysing the diversity of
reasons for which an utterance might be conversationally inappropriate, Grice demonstrated the
limitations to what can be concluded from such data. But his theory of conversation was not just a
factor in the implosion of ordinary language philosophy; it has a far more lasting and positive value.
It is massively cited by linguists, because it is the starting-point for much contemporary work in
pragmatics. But it also continues to play a vital negative role in contemporary analytic philosophy.
Analytic epistemology provides a good case study of the philosophical application of Gricean
pragmatics outside the philosophy of language. Analytic epistemologists today typically regard the
object of their study as knowing (or justified believing) itself, as opposed to the corresponding words
or concepts. In reflecting on knowledge or justified belief, they work through example after example
of epistemologically suggestive situations, often of quite everyday sorts. In determining how to
describe such situations, they frequently have to ask themselves whether a proposed description is
false or, by contrast, true but conversationally misleading because it has a false conversational
implicature. They use Grices theory of conversation to filter out contaminated data. They also have
to engage with semantics as well as pragmatics, since some of the leading contender theories are
contextualist, in the sense that they postulate shifts in the reference of epistemic terms according to
the context in which they are used. Despite all that, the epistemologists underlying object of study
is knowing itself, not the verb to know or the concept of knowing. They sound like ordinary
language philosophers, and in a loose enough sense they are ordinary language philosophers, even
when ordinary language plays no special role in their epistemological aims (Hawthorne 2004 and
Stanley 2005 are good examples). In that respect, ordinary language has returned to its origins. For
the classic manifesto for ordinary language philosophy, at least in its Oxford form, appeared in J.L.
Austins paper A plea for excuses (1956-7). But much of Austins discussion of philosophical method

15
there appears strongly influenced by similar comments in the work of John Cook Wilson, Wykeham
Professor of Logic at Oxford from 1889 until his death in 1915, founder of an Oxford tradition of
realist metaphysics and knowledge-centred epistemology, and by no means a linguistic philosopher
as the term is usually understood. For instance, Cook Wilsons remark Distinctions current in
language can never be safely neglected (1926, p. 46) is echoed in Austins emphasis on the value of
starting with distinctions robust enough to have survived in ordinary language. Cook Wilsons ideas
and writings still loomed large in the Oxford philosophy of Austins student days, not least through
the influence of his star pupil H. A. Prichard, the Whites Professor of Moral Philosophy from 1927 to
1937, whose lectures Austin attended as an undergraduate despite Prichards attempt to ban him
for asking too many questions. Thus many contemporary analytic philosophers pay close attention
to linguistic subtleties, without treating their primary subject matter as in any way linguistic.
The foregoing sketch suggests no easy moral, except that the closer one looks at the history
of anything, the less it lends itself to easy morals. But I can add some colour to the picture, and
another perspective on the transition, by going back to my own experience of it at Oxford.

II

I arrived as an undergraduate at Oxford in 1973, to study Mathematics and Philosophy. Logic played
a central role in the course, to my lasting benefit; its centrality is relevant to the point of view from
which I observed the scene. I received my undergraduate degree in 1976 and began to study for a
doctorate. Originally I hoped to formalize Leibnizs principle of sufficient reason, but I soon switched
to Karl Poppers idea of verisimilitude, on which science can progress through a succession of
theories that get closer and closer to the truth without ever quite reaching it. I left Oxford in 1980, to

16
start my first proper job, as a lecturer in philosophy at Trinity College Dublin, and received my
doctorate in 1981.
In 1973, the two senior professors of theoretical philosophy at Oxford were A. J. (Freddie)
Ayer, Wykeham Professor of Logic from 1959 to 1978, and P. F. (Peter) Strawson, Waynflete
Professor of Metaphysics from 1968 to 1987. Austin had died in 1960 and Prior in 1969, both
prematurely; Grice had left for Berkeley in 1967. One could schematically associate Ayer and
Strawson with the two main strands of mid-century analytic philosophy: logical positivism and
ordinary language philosophy respectively. Ayer had studied with the Vienna Circle and his first
book, Language, Truth and Logic (1936) contained much logical positivist doctrine, including a
critique of metaphysics based on the verification principle, although he traced his genealogy further
back: The views which are put forward in this treatise derive from the doctrines of Bertrand Russell
and Wittgenstein, which are themselves the logical outcome of the empiricism of Berkeley and David
Hume (1936, p. 31). The book had been notorious for its advocacy of expressivism about morality
and religion. Its cheeky, provocative style is conveyed by the title of the last chapter, Solutions of
outstanding philosophical problems. Whereas Ayer was a follower of Russell, Strawsons most
famous article (1950) was a critique of Russells prize contribution to philosophy, his theory of
descriptions, and Strawsons first book argued more generally that such applications of modern logic
did no justice to the subtleties of ordinary language (1952). In live discussion, Ayer used rapid fire,
Strawson elegant rapier play. Ayer was better known to the general public, as a radio personality;
Strawson was more highly rated by professional philosophers, as more original. Strawson had been a
candidate for the Wykeham Chair of Logic; when Ayer was elected, through the votes of the nonphilosophers on the committee, Austin and Ryle resigned in protest. Asked that evening by a
colleague whether he felt very disappointed not to have been elected, Strawson replied Not
disappointed, just unappointed.

17
By 1973, it was no longer strictly appropriate to classify Ayer as a logical positivist, or
Strawson as an ordinary language philosopher. It was closer to the mark to describe Ayer as a
Humean, and Strawson as a Kantian: the contrast between them was no less marked for that.
Strawsons development into a systematic metaphysician has already been noted. Ayer had
retreated from his early radicalism, including the verificationist critique of metaphysics. He
commented that the trouble with Language, Truth and Logic was that all its main doctrines were
false. In 1976, on the fortieth anniversary of its first publication, he gave a series of lectures about
what remained of his original view. In the book, he had quoted as an example of an unverifiable
pseudo-proposition of metaphysics the Absolute enters into, but is itself incapable of, evolution and
progress, which he describes as A remark taken at random from Appearance and Reality, by F. H.
Bradley (1936, p. 36). In his lecture, he admitted that, far from having taken the remark at random,
he had searched through Appearance and Reality for hours to find something suitably nonsensicalsounding. It is a salutary reminder of the intelligibility by ordinary standards of much metaphysical
discourse especially when not torn out of context in the way he presented the passage from
Bradley.
To younger philosophers in 1973, Ayer appeared quite old-fashioned philosophically. So too,
though to a lesser extent, did Strawson. The underlying reason was in large part their relation to
modern formal logic in philosophy. Officially, Ayer was for it and Strawson against it, but neither of
them knew much about it. They had received their philosophical education at a time when such logic
did not loom large in Oxford. Philosophers of that generation sometimes referred to formal logic as
sums, the primary school word for elementary arithmetic. The effect of Ayer and Strawsons lack of
facility with modern formal logic was that they were poorly placed to deal with the new wave of
philosophy of language sweeping across the Atlantic, led by Kripke and Lewis (who was at least as
much a philosopher of language as a metaphysican in the 1970s), and other philosophers and
linguists such as Donald Davidson, Hilary Putnam, David Kaplan, Robert Stalnaker, Keith Donnellan,

18
Richard Montague, and Barbara Hall Partee. New wave philosophy of language involved the
application of formal semantics, based on modern logic, to natural languages.
Ayer had never had a detailed interest in the philosophy of language. His resentment of the
new wave focussed on Kripkes case for the necessary a posteriori and the contingent a priori, in
effect because those categories violated Humes supposedly exhaustive distinction between matters
of fact (contingent and a posteriori) and relations of ideas (necessary and a priori). An annual ritual
took place in Ayers Informal Instruction, his class open to all comers, where a short presentation of
some recently published work would kick off the discussion. Many of the brightest graduate
students attended, even though they had joined the new wave, for Ayer was good at creating an
atmosphere conducive to discussion. But every year he would read a short paper purporting to
refute Kripke on the necessary a posteriori and the contingent a priori, in fact based on exactly the
confusions Kripke had done so much to clear up. When he had finished, the graduate students
would by implication plead with Ayer not to misunderstand Kripke in vain. Strawson was much
more of a philosopher of language than Ayer, but even his perception of new wave philosophy of
language was distorted by the old-fashioned lens of an exaggerated contrast between, in effect,
ordinary language philosophy attentive to speakers actual use of natural language in all its
complexity and ideal language philosophy trying to project the simple logical structure of a formal
language onto natural language, in abstraction from its speakers, with Procrustean effect (Strawson
1971). What he never properly appreciated was the new wave conception of the two projects as
mutually complementary rather than in competition, so that interpreting a natural language in terms
of a comparatively simple formal truth-conditional semantics would make the best sense of the
complexities of speakers actual use of the language (Lewis 1975). That Strawsons criticisms of newwave philosophy of language were widely felt to miss the point contributed to his looking like a
figure from the past too.

19
In effect, new wave philosophy of language achieved a surprising reconciliation between
elements from the two main competing strands of mid-twentieth century analytic philosophy, logical
positivism and ordinary language philosophy. From logical positivism it took the rigorous use of
formal languages with precisely and systematically described syntax and semantics, as found in
modern logic, to model meaning. From ordinary language philosophy it took most of its data about
use to be explained, as well as ideas about the nature of the relation between meaning and use, in
order to bring the formal models to bear on the data. An encouraging precedent was Noam
Chomskys success in explaining subtle, apparently messy complexities in the surface syntax of
English in terms of formal models of a postulated underlying deep structure (Chomsky 1957). What
new wave philosophers of language hoped to do for the semantics of natural languages seemed
analogous to what Chomsky and others were already doing for syntax despite Chomskys own
scepticism about the scientific status of semantics. Indeed, it was natural to expect a tight relation
between the semantic and syntactic structure of an expression, at least at the level of deep structure
or logical form. For it was a fundamental tenet of new wave philosophy of language, coming through
Carnap from Gottlob Frege, that the semantics must be compositional, in the sense that the meaning
of a complex expression is determined by the meanings of its constituents; how else to explain our
ability to understand sentences we have never previously encountered, if made up of familiar words
in familiar types of combination? The initial hypothesis must surely be that the requisite semantic
articulation of sentences into their semantic constituents matches their syntactic articulation into
syntactic constituents at some deep enough level. The compositionality constraint exerted a
powerful force in the direction of systematicity. In practice, the only semantic theories to exhibit
(rather than merely claim) such a compositional structure were those for formally specified
languages. Without such a formal semantics, a philosophy of language looked badly undeveloped to
new wavers. It was partly for this reason that Austin was barely mentioned in Oxford philosophy of
language by 1973, since he had no formal semantics to offer. The same went for his protg and in
some respects intellectual heir John Searle, whose major work Speech Acts looked out of date from

20
an Oxford perspective almost as soon as it was published in 1969. Although Austin and Searle
continued to exert a significant influence, it was mainly outside the new wave.
New wave semantics came in two main varieties, though methodologically the differences
between them were minor compared to their shared differences from their predecessors. One
variety was possible worlds semantics, which went back to Carnap and by 1973 was associated with
philosophers of language such as Kripke, Lewis, Kaplan, Stalnaker, and Montague. In English as a
formal language (1970), The proper treatment of quantification in ordinary English (1973), and
other papers, Montague showed how it could provide a rigorously working compositional semantics
for large fragments of a natural language. His work had a major influence on Barbara Hall Partee and
has been seminal for a major tradition of intensional semantics as a branch of linguistics. The other
main variety of new wave semantics was extensional, under the influence of Quine. Chastened by his
scepticism about meanings, it approached the semantic realm less directly, through theorizing
explicitly about reference and truth rather than meaning itself. Nevertheless, its emphasis on the
constraint of compositionality was just as strong. Formally, it took inspiration from Tarskis theory of
truth. Its main proponent was Donald Davidson (1967a). It too had a significant impact on linguistics,
most notably through Davidsons semantics for verbs and adverbs in natural language, which
postulated tacit quantifiers over events (Davidson 1967b). A leading Davidsonian linguist was James
Higginbotham.
I first encountered new wave philosophy of language in my first term as an undergraduate,
when my tutor encouraged me to attend the John Locke lectures, to be given by the rising young
star of American philosophy, Saul Kripke. I was hugely impressed by his clarity, informal rigour,
pointed examples, common sense, and humour. The lectures were mostly non-technical, but one
both sensed and independently knew of his easy technical mastery of the subject. Although I did not
think of it this way at the time, Kripke combined and reconciled the virtues of ideal language
philosophy with those of ordinary language philosophy. At that stage, of course, I knew very little of

21
the background in the philosophy of language to what he was saying, and was in no position to
follow everything that went on in the lectures and the discussion that followed them. Nevertheless,
Kripke became the nearest I had to a model of how to do philosophy.
Although Kripkes work was widely discussed in Oxford at that time, especially by younger
philosophers, the dominant variety in town of new wave semantics was extensional rather than
intensional. Davidson had given the John Locke lectures in 1970. The two most admired young
theoretical philosophers in Oxford, Gareth Evans and John McDowell, somehow combined Davidson
with Frege as the packed audiences at their joint classes looked on. It was the moment of the
Davidsonic boom although people tended to utter the Tarskian mantra The sentence Snow is
white is true in English if and only if snow is white in a rather slow and quiet voice. If you wanted to
write on the philosophy of X, whatever X was, then you were supposed to start by writing a truth
theory for a language for talking about X. I found the atmosphere of reverence for Davidson
unhealthy, not to say sickening. It was fine for him to choose a speculative and controversial
extensionalist starting point for his programme for the philosophy of language, and to take it as far
as he could from there, but the project cried out to be undertaken in a scientific spirit, as a test of its
assumptions. Instead, they were treated especially by those lower in the hierarchy as dogmas
of mysterious but compelling power, an attitude encouraged by Davidsons elliptical and slightly
evasive style. It was recognized that Davidsons programme had to meet the challenge of providing a
compositional semantics for various apparently non-extensional constructions in natural language,
for ascribing beliefs and desires or possibility and necessity, for example, but the discussion of
alternative proposals muffled the issues with a lack of openness and clarity about the rules of the
game. Philosophers, some with only a rather tenuous grasp of technical matters, would invoke
obscurely motivated technical constraints a ban on substitutional quantification, say, or a
requirement of finite axiomatizability to exclude rival hypotheses. Kripkes critique of
Davidsonians objections to substitutional quantification (1976) has been condemned as cruel, but to
me it came as a breath of fresh air; I can attest to the presence at the time of the sort of atmosphere

22
about which he complained.2 By contrast, intensional semantics seemed to be conducted in a more
open, scientific spirit, though Davidsonians objected, darkly, to its possible worlds as creatures of
darkness.
Davidsonians did not expect the philosophy of language to be independent of metaphysics.
Davidson (1977) explicitly motivated an ontology of events by a Quine-inspired principle of charity in
interpretation, through the semantics of adverbs. This was not so different from Lewiss original
motivation of his ontology of possible worlds, in effect by the principle of charity, through the
semantics of modal operators. But Davidson gave his metaphysics a turn reminiscent of Strawson,
with a transcendental argument to show that it was not really possible to think differently (Davidson
1973-74). As so often with transcendental arguments, it turns out to depend on concealed
verificationist assumptions (see the remarks below on Wittgensteins Private Language Argument).
More recent metaphysics has been much less tempted by transcendental arguments.
Amongst the important features shared by the extensional and intensional varieties of new
wave semantics was truth-conditionality: they both treated the meaning of a declarative sentence as
in some sense the condition for it to be true. By contrast, the senior Oxford philosopher properly to
engage new wave philosophy of language, Michael Dummett, opposed such truth-conditional
semantics in favour of assertibility-conditional semantics, which treated the meaning of a declarative
sentence as the condition for it to be assertible, or verifiable, rather than true. Assertibilityconditional semantics was inspired by the proof-conditional semantics for mathematical language
developed by Heyting, Prawitz, and other intuitionists, which equated the meaning of a sentence in
the language of mathematics with the condition for something to be a proof of it. The plan was to
generalize this semantics to the whole of language by treating mathematical proof as a special case
of verification. This may be seen as a descendant of a logical positivist conception of the meaning of
a sentence as its method of verification, although Dummett did not present it as such.

23
Unlike the logical positivists, Dummett saw the subversive threat that verification-centred
semantics posed to classical logic, given the compositionality constraint which he accepted, in line
with new wave philosophy of language. For example, the natural compositional semantic clause for
disjunctive sentences of the form A or B says that A or B is verified if and only if either A is
verified or B is verified. But that makes immediate trouble for the classical law of excluded middle,
A or not A. For the special case of the semantic clause where B = not A says that A or not A is
verified if and only if either A is verified or not A is verified. But often we cannot verify a sentence
and cannot verify its negation. For instance, we cannot verify Napoleon had an even number of
hairs at his death, nor can we verify Napoleon had an odd number of hairs at his death. Thus, by
the semantic clause, we cannot verify Napoleon had an odd or even number of hairs at his death.
In these ways, Dummett saw the philosophy of language as calling into question a realist
metaphysical conception of reality as how it is quite independently of our capacity to find out how it
is, and pointing towards an alternative anti-realist metaphysics. In his view, the role of the
philosophy of language here is not merely evidential, to give us reasons to believe one metaphysical
theory and not another. Rather, he saw alternative theories of meaning as giving something like the
cash-value of alternative metaphysical pictures. Methodologically, he proposed to replace futile
quarrels between metaphysical pictures by comparisons between the corresponding theories of
meaning as the only scientific way to resolve the issue. For him, metaphysical disputes are not
senseless; nor are they what they seem, since they are implicitly disputes in the philosophy of
language. One might compare Dummetts understanding of metaphysics with Strawsons. For both
of them, metaphysical questions turn into questions about the structure and limits of coherent
thought, to be answered by systematic inquiry. But Dummett was more open to revisionary
metaphysics than Strawson was, because he took seriously the danger of fundamental incoherence
in our current ways of thinking (as with our acceptance of the law of excluded middle), which
Strawson did not. Dummett also differed from Strawson in regarding modern logic as a decisive
advance over its predecessors, including its capacity to provide formal methods and model formal

24
languages for use in the philosophy of language: he was adept himself in such methods. In that
sense too, he was a new wave philosopher of language.
Dummett laid out his programme early on in his career (1959). The year Kripke gave the
John Locke lectures, Dummett brought out his first magnum opus, Frege: Philosophy of Language
(1973), in which he engaged creatively with Frege to develop his own views. Central to Oxford
philosophy in my student days was the dispute between, on one side, realism and truth-conditional
semantics, represented by the Davidsonians, and, on the other side, anti-realism and assertibilityconditional semantics, represented by Dummett. My sympathies were strongly with realism, though
not with Davidsonianism. Dummett supervised me for the last year of my doctoral studies (1979-80),
at the start of his period as Wykeham Professor of Logic (1979-92), the first holder of that chair with
a deep knowledge of modern logic. He was remarkably tolerant of the strident realism of my thesis,
which effectively presupposed the futility of his lifes work and pursued other issues from that
starting-point. I cannot resist a couple of memories from that period.
When I told other philosophers at Oxford that I was working on the idea of approximation to
the truth as applied to scientific theories, their reaction was always to ask Is that something to do
with vagueness? For vagueness was a big issue in Oxford then, being conceived as a major
challenge to realism, truth-conditional semantics, and other forms of orthodoxy (Dummett 1975,
Fine 1975, Wright 1975). I always found that reaction annoying, because I thought it betrayed a
myopic obsession with the philosophy of language. I would reply that my thesis had nothing at all to
do with vagueness, pointing out that, of two perfectly precise but false scientific theories, one may
be a better approximation to the truth than the other (I also enjoyed shocking people by saying that
I found Popper more interesting than Davidson). On the narrow issue I was right, but my later
trajectory suggests that my interlocutors were not completely wrong about the direction of my
interests (Williamson 1994). Indeed, one of my main reasons for later working on vagueness was

25
that it was generally regarded as a paradigm of a phenomenon in need of anti-realist treatment. I
wanted to strike at what was supposed to be the safest fortress of anti-realism.
Another memory comes from a supervision with Dummett. We were discussing an argument
that I thought one of the best in my thesis, and he thought one of the worst; it later became the
kernel of the only publication that emerged from my doctoral studies (Williamson 1988). After a
while, Dummett reflected and said The difference between us is that you think that inference to the
best explanation is a legitimate method of argument in philosophy, and I dont. I realized that his
characterization of the difference was right, although I was a little shocked at his outright rejection
of inference to the best explanation in philosophy. His view was something like this: the deep
philosophical issue will be about which of the theories that yield the putative explanations is so
much as meaningful; that issue must be settled first before we can judge the value of those putative
explanations; but once it has been settled, nothing much is left for inference to the best explanation
to do. I still favour inference to the best explanation and an abductive methodology in philosophy
(Williamson 2013, pp. 423-9). Indeed, it is hard to see how the kind of positive, systematic, general
theory that Dummett sought in the philosophy of language could be established by any other means.
He was optimistic about the long-run prospects of settling philosophical issues in a decisive,
systematic way, just as he took to happen in science (Dummett 1978). But if one tries to establish
the meaningfulness of a theory by an argument more decisive than abduction, wont one have to
first establish the meaningfulness of that argument by a further argument more decisive than
abduction? There starts an infinite regress. Dummetts dislike of inference to the best explanation
may help to explain why his discussion of assertibility-conditional semantics never really got beyond
the programmatic stage to the nitty-gritty of properly developing models of such semantics for nontrivial fragments of non-mathematical language. Even if such models had worked well (a tall order),
they would at best have provided him with some sort of abductive argument in favour of his
programme, whereas he wanted something more decisive. In the long run, the failure of his
programme to develop such working models has been a major reason for its marginalization,

26
especially when combined with its radically revisionary and implausible consequences for logic and
metaphysics.
Dummett presented his views on the relationship between the philosophy of language and
metaphysics in his William James Lectures at Harvard in 1976, the same year that Hilary Putnam
gave the John Locke Lectures at Oxford. A book soon grew out of Putnams lectures, showing signs of
Dummetts influence in some rather unfortunate arguments against something called metaphysical
realism (Putnam 1978). But it was not long before the anti-realist Dummett was replaced by the
realist (if not metaphysical realist) J.L. Austin as the main Oxford presence in Putnams work (Putnam
1994). Dummetts lectures took much longer to appear in print, as The Logical Basis of Metaphysics
(Dummett 1991). For a key work of a major philosopher, it had comparatively little impact. One
problem was that its discussion of proof theory as the foundation of semantics was informal, often
elliptical, digressive, or vague, with philosophical and purely technical matters all mixed together,
making it excessively and unnecessarily hard for logicians to extract and perhaps answer the purely
technical questions raised. Moreover, the generalization of the semantics to non-mathematical
language still remained at a tentative, programmatic stage, not conducive to applications in
linguistics. A more general problem for Dummett was that by 1991 the philosophical zeitgeist was
even less receptive to anything like Dummetts programme than it had been earlier. He did not
engage with the new paradigms of metaphysics, such as the work of David Lewis. There was no
obvious way to interpret the new metaphysical theories as picturesque guises for views in the theory
of meaning, nor did the new generation of metaphysicians wish to do so. Metaphysics itself had
grown in self-confidence and felt no need to present itself as anything else. Incidentally, despite its
title, my own book Modal Logic as Metaphysics (2013) is very far from a return to Dummetts
understanding of the relationship between logic and metaphysics. As a first approximation, logic in
Dummetts title means something like philosophical reflection on the meaning of the logical
constants, while in mine it means generalizing about the world in terms just of the logical
constants. For Dummett, logic is metalinguistic, for me it is not.

27
By the 1990s, few readers felt in danger of being compelled, against their wills, by
Dummetts convoluted arguments, even when they understood them. One of several reasons was
that he discussed the mind in ways that still carried behaviourist baggage from a philosophy of mind
widely rejected by younger generations since the collapse of behaviourist psychology and its
replacement by cognitive psychology in the 1960s. Such baggage is detectable in remarks like this
about knowledge of meaning: we should [] not be content with saying what is known, without
saying what it is to have that knowledge, that is, how it is manifested by one who has it (Dummett
1991, pp. 104-5), where the manifestation is in observable behaviour. In this respect, Dummett can
be compared to Quine, who was influenced by his Harvard colleague Skinners behaviourism in
psychology. One might think that behaviourism about language had been outdated since the
publication of Chomskys famously destructive review of Skinners Linguistic Behavior (1959), but
digestion can be a slow process. I remember sophisticated young philosophers of language at Oxford
in the late 1970s still talking of children learning their native language by being drilled in it by
adults. In Dummetts case, his behaviourist tendencies came from his reading of Wittgenstein rather
than Skinner, and were correspondingly subtler and less eliminativist than Quines. Nevertheless, the
differences should not be exaggerated. An anecdote from late in Dummetts career: A group of
younger Oxford philosophers were discussing what he meant, in the great mans silent presence.
Suggestion after suggestion was rejected because it would attribute to him crude old-fashioned
behaviourism. Eventually, someone turned to him and asked So what is your view, Michael?.
Dummett replied I think its the one youve been calling crude old-fashioned behaviourism.
For Dummett, as for other British philosophers of his generation, Wittgensteins central
contribution to the philosophy of mind was his Private Language Argument. Its interpretation was
disputed, but it was widely supposed to show something very deep about the need for talk about
mental states to involve observable criteria (in some sense) for attributing them to others. The
putative insight had widespread repercussions for the philosophy of language, concerning not just
the semantics of mental state ascriptions but the nature of the mental states in play for speakers

28
and hearers of any speech act, and in particular the nature of understanding. Dummetts preference
for assertibility-conditions over truth-conditions in the theory of meaning was rooted in the close
linkage of assertibility-conditions to the observable use of the language, which realist truthconditions lacked, since users of the language might have no idea whether they obtained. He
combined this Wittgenstein-inspired focus on use with a Frege-inspired insistence on the need for a
systematic, compositional theory of meaning, modelled on the semantics of a formal language. In
thus uniting elements of the two previous traditions of analytic philosophy, ordinary language
philosophy and ideal language philosophy, Dummett resembled the younger new wave philosophers
of language, although his selection of elements to combine differed from theirs.
At this point something must be said more generally about the influence of Wittgenstein on
British philosophy in the period under discussion as has often been remarked, his influence in
North America was never as great as in Europe, one reason being the greater sway of naturalism or
scientism in North America, led by Quine and others. In the case of Dummett, since he was a student
at Oxford in 1950 when Wittgenstein spent some time there living in Elizabeth Anscombes house,
future historians might wonder whether there was face-to-face influence. It is therefore worth
recounting the story Dummett liked to tell about his only meeting with Wittgenstein. Dummett was
going for a tutorial at Anscombes house. She kept the door unlocked. As was the practice, Dummett
went in, and sat down to await her summons. An elderly man in a dressing-gown came downstairs
and asked Wheres the milk?; Dummett replied Dont ask me. That was the extent of his
conversation with Wittgenstein. What mattered instead was Anscombes mediating role. She was
probably the strongest transmitter of Wittgensteins influence at Oxford until she left in 1970 to take
up a chair at Cambridge, although of course she was always a fiercely independent-minded
philosopher in her own right. Other Oxford ordinary language philosophers such as Ryle, Austin, and
Grice were not moulded by Wittgenstein, and the number of card-carrying Wittgensteinians at
Oxford was never very high. Nevertheless, his influence was still pervasive when I was a student in
the 1970s. Gordon Baker and Peter Hacker, guardians of the flame, had a large following amongst

29
graduate students. They were later to have an ill-tempered dispute about the value of Freges
philosophy with Dummett: when their disparaging book on Frege was published, Dummett
organized an emergency series of graduate classes to denounce it (Baker and Hacker 1984, Dummett
1984). It is easy to list many Oxford philosophers of the time whose work showed significant
Wittgensteinian influence to varying extents, even though it would be crass to classify them simply
as Wittgensteinians: Dummett, Strawson, Philippa Foot, Iris Murdoch, David Pears, Anthony Kenny,
of a younger generation John McDowell and Crispin Wright, and so on. What may be less obvious is
how wary even those who barely mentioned him were of plainly saying that he was wrong about
something. One knew that doing so incurred the automatic charge of shallow misinterpretation. It
was best to step quietly around, and let sleeping dogs lie.
Wittgensteins main influence at that time was through his later work, although few of those
under the influence imitated his style of philosophizing in that work. Most engaged in overt
theorizing of a more or less systematic kind. The citadel was the Private Language Argument, from
which he exerted his power over the philosophy of mind and the philosophy of language. The
growing external threat to that power from cognitive psychology was surprisingly little felt in 1970s
British philosophy. But there was also an internal threat. For how exactly was the Private Language
Argument supposed to work? Wittgensteins presentation was notoriously Delphic. The simplest and
clearest reconstructions had the argument rest on a verificationist premise to the effect that one
couldnt be in a mental state unless some independent check was possible on whether one was in
that state. But it was generally agreed that if the argument rested on a verificationist premise then it
was not compelling, because verificationism could not just be assumed without argument.
Defenders of the argument insisted that it worked without such a premise, but could not
satisfactorily explain how (a similarity with Davidsons transcendental argument mentioned above).
Wittgensteins citadel was in danger from within; his power was waning as a result. At this point an
unlikely would-be rescuer arrived: Saul Kripke. In lectures from 1976 onwards, and in his book on the
Private Language Argument and the associated considerations on rule-following (Kripke 1982), he

30
offered a conjectural interpretation of the argument that was clearly non-verificationist and, if not
compelling, at least powerful. The question was: did it fit what Wittgenstein meant? The consensus
amongst Wittgensteinians was that it did not, and as a matter of historical scholarship they may well
have been right. But they seemed not to realize that in taking the negative attitude they did, they
were also rejecting their last chance to avoid marginalization from the philosophical mainstream.
The power of Wittgensteins name resumed its decline. As for Kripkes argument in its own right, it
inadvertently gave the new metaphysics an opportunity to spread its influence. For Kripkes
argument took the form of a sceptical paradox, to which Kripke offered a rather unclear and
unattractive radically sceptical solution. By contrast, David Lewis offered a clearer and more
attractive non-sceptical solution, by means of a highly metaphysical distinction between objectively
natural and objectively non-natural properties (Lewis 1983b). It was something like Lewiss solution,
not Kripkes, that was widely accepted.
Here are two snapshots of the decline in Wittgensteins standing. The first is of a meeting in
about 1994 of the Tuesday group, originally founded by Ayer on his return to Oxford in 1959 as a
counterweight to Austins Saturday morning meetings. Susan Hurley read a carefully reasoned paper
against the Private Language Argument to an audience that included many leading Oxford
philosophers. The audience divided by age. Roughly, those over fifty did not take the possibility
seriously that Wittgensteins argument was fundamentally flawed, although they also did not explain
how it worked or what it showed; those under fifty were more sympathetic to Hurleys objections.
The second snapshot is of a large graduate class on philosophical logic shortly after my return to
Oxford in 2000. One student kept pressing the Wittgensteinian line that contradictions are
meaningless rather than false. I kept giving the standard responses, that contradictions have true
negations while the negation of what is meaningless is itself meaningless so not true, that the
compositional semantics generates meanings even for contradictions, and so on, whose effect was
merely to elicit variations on the same theme that did not meet the objection. Eventually I became
exasperated and said Maybe Wittgenstein was just wrong; it wouldnt be the first time. There was a

31
collective gasp of shock. I have never again witnessed such a reaction when Wittgensteins name
was taken lightly.
Of course, the flame is kept alive by surviving groups of Old Believers. Some others, more
willing to believe that there has been progress in philosophy since 1970, still find value in engaging
with Wittgensteins work. Nevertheless, his influenced has declined drastically over the past forty
years. No doubt that could be roughly measured by his proportion of citations in journals. But what
strikes me most forcefully is that the fear factor has gone. As a test of authority, of intellectual or
other kinds, admiration tells less than fear. In the 1970s, even non-Wittgensteinian philosophers
were often afraid to speak out against Wittgenstein. They are so no longer. Another philosopher
who has ceased to elicit the fear factor is Quine. Originally, he was frightening because few could
match his skill with the weapons of formal logic in philosophical debate. By the 1970s that was no
longer so, but philosophers were still very nervous of relying on ordinary semantic notions such as
synonymy, because they were afraid of being caught out by Quines argument for the indeterminacy
of translation. That fear too gradually evaporated in the 1970s, as Quines behaviourist assumptions
fell into disrepute.
Having said so much about the Oxford philosophical scene in the 1970s, I should continue
the story into the 1980s and 90s. It was not at all a linear extrapolation from the 1970s. Strikingly,
new wave philosophy of language receded fast (though it proved temporarily) in Oxford, less so
elsewhere. One reason was tragically extrinsic: Gareth Evans died at the age of 34 in 1980. With him,
the new wave in Oxford lost much of its technical panache, and detailed work in semantics dropped
off. Although James Higginbotham was Professor of General Linguistics from 1993 to 2000, not much
else was going on at Oxford in Davidsonian semantics for him to engage with. The Davidsonic boom
had come to an odd end, morphing into moral philosophy in the work of John McDowell, David
Wiggins (Wykeham Professor of Logic from 1994 to 2000, in succession to Dummett), Mark Platts,
and others. The transition was made through the Davidsonian emphasis on the legitimacy of

32
homophonic truth theories, in which a word is used to state its own reference. Contrary to
appearances, it is not trivial that round in English applies to all and only round things, because one
has that to learn in learning English. Semantic analysis cannot go on for ever; eventually we reach
semantic atoms, and switching to a non-homophonic semantics achieves nothing to the purpose,
because the aim of semantics is not to write a textbook that one might read in order to learn the
object-language from scratch, but rather to say explicitly what its expressions mean in a systematic,
compositional way, to those who may already understand them implicitly. Those in the intensionalist
strand of new wave philosophy of language had to resort to homophonic lexical semantics too. The
Davidsonians realized that, in particular, they could just as well give a homophonic semantics for
moral language too (Wiggins 1976). For instance, evil in English applies to all and only evil things.
Nothing in their philosophy of language made it problematic to give such an ostensibly out-and-out
realist treatment of moral language. Nor did it require any further semantic analysis of moral terms;
they could be treated as unanalysable. Thus Davidsonian philosophy of language found itself in the
unaccustomed role of providing a protective environment for Aristotelian moral realism. By contrast,
Dummett put much heavier explanatory demands on the theory of meaning, perhaps too heavy to
be satisfiable.
From the late 1970s onwards, Dummett also found himself fighting a more global trend in
analytic philosophy: a move away from the philosophy of language towards the philosophy of mind.
On his picture of the history of philosophy, Descartes had made epistemology first philosophy, the
engine for the rest of philosophy, and then Frege had replaced epistemology by the philosophy of
language as first philosophy. Analytic philosophy was philosophy downstream from that linguistic
turn. But many analytic or ex-analytic philosophers were starting, heretically, to treat the philosophy
of mind as more fundamental than the philosophy of language. Cognitive psychology made a far
more interesting and attractive conversation partner for philosophy than behaviourist psychology
had done, and had a natural interface directly with the philosophy of mind, for instance in the theory
of perception. Computer models of the internal workings of the mind were also increasingly

33
influential. Once again, many of the innovations came from North America. As behaviourism lost its
authority, Thomas Nagel (1974) led the way in talking directly about conscious experience. Although
Daniel Dennett (1981) still showed some influence from his Oxford supervisor Ryle, he engaged with
psychology through the philosophy of mind, not the philosophy of language. Jerry Fodor (1975)
postulated a language of thought, on the model of a computers machine code, but it was to be
studied by the methods of psychology and computer science, not those of linguistics. Moreover, it
was not a public language, whereas Dummett envisaged first philosophy as the philosophy of public
language, in line with the Private Language Argument.
In Oxford, the move into the philosophy of mind took a very specific form, which Dummett
had unintentionally facilitated. For him, much of Freges achievement in the philosophy of language
depended on his distinction between sense and reference. Sense is individuated cognitively: two
senses may present the same reference in ways which count as different because they fail to render
the sameness of reference transparent to the thinker. The cognitive nature of sense promised to
make the connection Dummett wanted between the semantics of a language and speakers use of
that language. Thus Hesperus and Phosphorus differ in sense and use but not in reference.
Dummett followed Frege in making sense a level of linguistic meaning distinct from the level of
reference. Initially, this gave Fregean semantics a large head start over one-level referential
semantics, like that associated with Russell, in explaining linguistic phenomena such as the apparent
difference in truth-conditions between Mary thinks that Hesperus is bright and Mary thinks that
Phosphorus is bright. However, new wave philosophy of language in North America tutned against
Frege, especially at the level of public language. In particular, the semantic property of a name that
speakers share is its reference; as Kripke and even Frege emphasized, the names cognitive
connections may vary wildly from one speaker to another. Something similar applies to terms whose
reference depends on context: the linguistic meaning of the phrase that dog does not encode the
rich cognitive connections that it will have when used as a perceptual demonstrative by a particular
speaker on a particular occasion. Most of the younger Oxford philosophers of language in the 1970s

34
and 1980s followed Dummett in his Fregean sympathies. But, more impressed than Dummett by the
work of Kripke and other North American new wave philosophers of language, they applied the
sense-reference distinction only at the level of individual users of the language, not at the level of
the language as a whole. If senses are cognitively individuated determinants of reference, a proper
name expresses different senses for different speakers, and a perceptual demonstrative expresses
different senses on different occasions even for the same speaker. If senses are structured, much of
that structure will be at the level of thought and not at the level of language. Sense was often
glossed as a way of thinking of the referent. This shift in focus from language to thought was
already visible in the work of Evans (1982). By Dummetts standard, it meant that Evans and the
others who took that turn no longer even counted as analytic philosophers.
Thus Dummett found himself fighting on the home front too, trying to reassert the primacy
of language over thought in philosophical method. Although he was happy to regard philosophy as
the study of thought of what is thought, not the act of thinking it he insisted that the proper
way for philosophers to study thought was by studying it as expressed in public language, which the
neo-Fregean philosophers of thought no longer did. Perhaps they were not in direct contravention of
the Private Language Argument, because their senses could in principle be shared. Nevertheless,
from Dummetts methodological perspective, they had taken a step backwards, because the study of
public language gave philosophy the objective discipline it needed. To replace that discipline by the
objective discipline of experimental psychology would be, from his perspective, to commit the
disastrous error of psychologism, against which Frege had railed: it would involve confusing what is
thought with the act of thinking.
Dummett seemed to be fighting a losing battle. Globally, the centre of gravity of analytic or
post-analytic philosophy moved towards the philosophy of mind in the 1980s. Logic and semantics
suffered a significant loss of prestige: graduate students became less convinced of the need,
intellectual or professional, to put in the hard work of learning them. Locally, neo-Fregean

35
philosophers of thought were taking over. For instance, Strawson was succeeded as Waynflete
Professor of Metaphysics by Christopher Peacocke, who held the post from 1988 to 2000. Senses
became concepts (Peacocke 1992).
Since the 1980s, the philosophy of mind worldwide has continued to enjoy a far more
fruitful relationship with experimental philosophy than it did in the heyday of behaviourism.
However, it did not become first philosophy in the way it had been expected to do. Nor did the
philosophy of thought, which anyway never solidified as a recognized branch of the subject. For
instance, developments in metaphysics have typically not been driven by anything in the philosophy
of mind. After all, with regained confidence in metaphysics, its contemporary practitioners tend to
see themselves as investigating the most general and fundamental nature of a world in which
human minds play only a very minor role. Why should the philosophy of mind or the study of
concepts drive metaphysics any more than it drives physics? In principle, even if it cannot contribute
towards constructive metaphysical theory-building, it might help towards understanding the folk
metaphysical beliefs that may obstruct our acceptance of the correct revisionary metaphysics. In
practice, the philosophy of mind and the study of concepts have had little impact on recent
mainstream metaphysics even in that modest negative way.
For the past several decades, no branch of philosophy has played the fully-fledged role of
first philosophy within analytic philosophy. To some extent, that reflects the increasing specialization
of academic research in general. But it also concerns a change more specific to analytic philosophy
(in a sense broader than Dummetts), in what philosophers take their subject matter to be. As
already noted, an increasingly prevalent, broadly realist attitude is that when you are doing the
philosophy of X, you are primarily interested in X itself, in its most general and fundamental aspects,
and only secondarily in the word X, or our concept of X, or our beliefs about X, or our knowledge of
X. You are not surreptitiously doing the philosophy of language or thought or mind or knowledge.
This reconception of the subject gives no branch of philosophy a head start over the others.

36
However, the situation is more complicated than those simple formulations suggest. For
they might lead one to expect the philosophy of language to be just one more branch of philosophy
alongside all the others, the philosophy of a phenomenon specific to humans and perhaps some
other species scattered here and there over the universe. It looked like that to some in the 1980s,
and taken in isolation it may still sometimes look like that. But, as already suggested, the philosophy
of language also plays a more general role throughout analytic philosophy, in the evaluation of
arguments. Of course, we do not need the philosophy of language to determine whether an
argument is deductively valid in simple cases. But on almost any view of philosophy, it often involves
arguments with a subtle illusion of validity, and other arguments that are really valid but need to be
checked for such subtle illusions. The illusion may come from confusions between entailments and
presuppositions or conversational or conventional implicatures, or from concealed shifts in context,
or from lexical or syntactic ambiguities, or from other linguistic complexities. Any discipline that uses
subtle, complex would-be deductive arguments in natural language about abstract issues is liable to
such illusions, and philosophy characteristically uses arguments of that kind. That is of course not to
say that it uses nothing else, or that no other discipline uses them at all; nevertheless, philosophical
methodology past and present may put more weight on such arguments than does the methodology
of any other discipline. The shift from a deductive to an abductive methodology makes less
difference here than one might have expected, because abduction involves the assessment of
amongst other factors the strength, explanatory power and consistency with the evidence of a
theory, which in turns depends on its deductive consequences. Thus simply using the methods of
analytic philosophy critically, by contemporary standards, takes some sophistication in both
semantics and pragmatics, irrespective of the subject matter under philosophical investigation. That
is a robust legacy from analytic philosophy of language for all philosophy.
One day, perhaps, cognitive psychology will have developed to a point at which it can be
usefully deployed to locate likely trouble spots for philosophical reasoning, for instance where
framing effects may be exerting an undue influence. Some experimental philosophers believe that

37
we have already reached that point. However, perhaps with a few limited exceptions, it is doubtful
that purely psychological methods have yet reached an adequate level of discrimination to be
usefully applied in the way that linguistic methods already can be. Just to be told that the order in
which material is presented can influence our judgment is of little help, since either we ignore the
material or it is presented in some order or other. For the time being, linguistics and the philosophy
of language offer more help than do psychology and the philosophy of mind when we check an
alleged deduction. In this limited respect, Dummett was right about the methodological danger of
assigning priority to thought over language, but not for the deep and permanent reasons he
envisaged.
For the evaluation of deductive arguments, the relevance of logic is even more obvious than
that of the philosophy of language. Of course, some would-be deductive arguments in philosophy
are cast in such seamlessly discursive form that no extant logical theory is of much use in evaluating
them. Nevertheless, in most branches of contemporary analytic philosophy complex would-be
deductive arguments often are articulated with sufficient clarity for formal logical skills to make a
significant difference to the reliability with which their validity is assessed. Thus logic makes an
instrumental contribution to philosophy in general similar in kind to that made by the philosophy of
language, and perhaps greater in degree.
The development of formal methods in recent philosophy has also extended the scope for
logic to make a more direct contribution to branches of philosophy usually conceived as other than
logic. For instance, in epistemology, models of epistemic logic enable us to work through the
consequences of epistemological claims in exactly described, appropriately simplified situations in a
far more rigorous and systematic way than would otherwise be available. Something similar goes for
decision theory too. The model-building methodology that has proved so successful in the natural
sciences can thereby be applied in philosophy too, and provides new insights into old problems. In
metaphysics, rival logics often supply powerful structural cores to rival metaphysical theories: for

38
instance, a quantified modal logic is the structural core of any properly developed theory of modal
metaphysics. Although not all of modal metaphysics is usefully treated as logic, a vital part of it is.
Logic, far from displacing metaphysics as the logical positivists hoped, is at its centre.
The history of philosophy makes a mockery of any limited vision of what philosophy is. It has
not followed the path laid out for it by the logical positivists, nor that laid out by the ordinary
language philosophy. Nor has i perhaps with a few limited exceptions t become a branch of
psychology, or of physics. Nevertheless, under all the surface turbulence, it somehow manages to
extract the residue it needs from each changing fashion. Who knows where the cunning of reason
will take it next?

III

History is often said to be written by the winners. In the case of analytic philosophy, however, there
is a danger that history will be written predominantly by the losers. One reason is that analytic
philosophy is a somewhat anti-historical tradition, especially where it most resembles a science, in
aspiration or achievement. For there it tends to be oriented towards the future rather than the past,
in the manner of a science hardly surprising when progress is expected. Those who do not like
history cannot complain when their history is written by people who are not like them. A second
reason is that recent analytic philosophy seems to subvert the global narratives it might otherwise
be tempting to tell about the history of the subject most notably, in the resurgence of realist
metaphysics, often unashamedly concerned with things in themselves. For those sympathetic to
Kant or Wittgenstein or Dewey, it must be tempting to see much recent analytic philosophy as an
insignificant anomaly, a passing throwback, in the long march of philosophy.

39
A case in point is Richard Rorty, who was admirably willing to step back, identify bold
patterns in the then-recent history of analytic philosophy, and list his heroes Kant, Hegel,
Wittgenstein, Dewey, Heidegger, Sellars, Brandom, ; no wonder his racy, deliberately provocative
stories have been so widely read. It is striking that the very large number of names of contemporary
philosophers villains as well as heroes in the index to Philosophy and the Mirror of Nature
(Rorty 1979) does not contain that of David Lewis, who had already published two highly muchdiscussed books and many articles, and been Rortys colleague at Princeton since 1970. Rortys radar
had missed a serious threat, the central figure in analytic philosophy for the coming decades. Rorty
was out of sympathy with most new wave philosophy of language, and the metaphysics that
increasingly accompanied it, because its referential approach to semantics came too close for his
comfort to making language a mirror of the world. For the future, he put his money instead on the
inferential approach, particularly in the neo-pragmatist form offered by Robert Brandom (1994),
focussed on the commitments and entitlements of speakers to make moves in the language game.
Brandom himself has his own grand narrative of the history of philosophy, in which tongue partly
but not wholly in cheek he presents himself as the natural successor to Kant and Hegel (Brandom
2009). But his inferentialism has remained at an even more programmatic stage than Dummetts,
lacking an equivalent of Dummetts connection with technical developments in proof theory by Dag
Prawitz and others. As a result, inferentialism has been far less fruitful than referentialism for
linguistics. In that crude sense, referentialism beats inferentialism by pragmatic standards.
Of course, we cannot expect a history of recent philosophy to remain neutral about the
future. Even the driest chronicle of who published what when has implicit standards of historical
significance in selecting whom and what to include. Good historical narratives discern patterns in
their material more explicitly and reflectively. This paper which manifestly does not aspire to the
depth or rigour of serious history has indicated a few of the messy complexities which any history
of recent analytic philosophy must try to order. Nevertheless, it does at least gesture towards some
larger patterns to be made explicit and reflected on. A chronicle is not enough.

40
The power of fashion in philosophy already ensures that its history will exhibit some
patterns, if only of the mob rushing here and there. Some of those fashions look foolish in
retrospect; most of them did at the time to non-sympathizers. But fashion is powerful in all
academic disciplines, even in mathematics for instance, concerning which branches of the subject
or styles of work carry most prestige. Nor is that merely an inevitable defect in any collective human
enterprise. Academic fashions arise because people trained in a discipline have some respect for the
judgment of others trained in the discipline as to what is good or fruitful work, worth imitating or
following up. When things go well, that mechanism enables the community to concentrate its
energies quickly where progress is being and will be made, to avoid wasted effort, and to raise
collective standards. It is a way of learning from others. The word fashion is most appropriate when
the level of deference to majority opinion becomes too high, stifling diversity and independence of
mind, making it harder in the long run for the community to back up out of a wrong turning, since it
loses its sense of the alternatives. But the rule of fashion is only an exaggerated form of something
no community can do without. Even the time and energy spent on bad ideas and misconceived
programmes has its value, since the effect of the investment is that their limitations are properly
explored and tested, so lessons are properly learnt. The history of academic fashions is the history of
how things once looked to highly intelligent and knowledgeable people.
The changes in philosophy discussed in this paper occurred in a period whose political,
social, and cultural history is already being written. Its philosophical history needs to be properly
written too, by historians of philosophy with at least enough sympathy for them to understand why
what so many philosophers did seemed a good idea at the time. There are encouraging signs that
such histories are just starting to be written. I look forward to reading them.

41
Notes

This paper derives from a talk given at the Faculty of Philosophy of Belgrade University in
September 2014. I thank Professor Miroslava Trajkovski for the suggestion that I might
depart from my usual practice and give a talk on an historical theme, based in part on my
own experience. Thanks to the audience for constructive discussion, and to Peter Vallentyne
for spotting some bad typos. Additional thanks go to Professor Slobodan Perovi for inviting
me to write up the paper for publication in this journal. He also pointed out the thematic
connection with Peter Strawsons 1977 lecture at Belgrade University on the nature of
analytic philosophy, subsequently translated into Serbo-Croat (as it then was) and published
in Theoria (Strawson 1977). On that visit to Yugoslavia, Strawson lectured in Belgrade,
Zagreb, and Sarajevo. He later commented: I registered a certain difference in atmosphere
in the three places. At least in academic circles the intellectual style seemed relatively
untrammelled in Belgrade and Zagreb, though the political tone was different. In Sarajevo,
where I was only allowed to give one of my two scheduled lectures and had minimal contact
with fellow academics, one perhaps time-serving young man in my audience suggested that
my lecture revealed an essentially bourgeois outlook. I replied But I am bourgeoisan
elitist liberal bourgeois. My interpreter commented, sotto voce, They envy you (Strawson
1998, p. 14). The contrast between Strawsons account of analytic philosophy and the
present one, almost forty years later, may be instructive.

My first article to be accepted for publication, though not my first to be published, protested
against the Davidsonian dogma that theories of truth qua theories of meaning had to be
finitely axiomatizable (Williamson 1984).

42
References

Austin, J. L. 1956-7: A plea for excuses, Proceedings of the Aristotelian Society, 57, pp. 1-30.
Ayer, A. J. 1936: Language, Truth and Logic. London: Victor Gollancz. Page references to the 2nd
edition, 1946.
Baker, Gordon, and Hacker, Peter 1984: Frege: Logical Excavations. Oxford: Blackwell.
Bradley, F. H. 1893: Appearance and Reality. London: Swan Sonnenschein.
Brandom, Robert 1994: Making it Explicit: Reasoning, Representing, and Discursive Commitment.
Cambridge, Mass.: Harvard University Press.
Brandom, Robert 2009: Animating ideas of idealism: a semantic sonata in Kant and Hegel, in his
Reason in Philosophy. Cambridge, Mass.: Harvard University Press.
Carnap, Rudolf 1947: Meaning and Necessity: A Study in Semantics and Modal Logic. Chicago:
University of Chicago Press.
Chomsky, Noam 1957: Syntactic Structures. The Hague: Mouton.
Chomsky, Noam 1959: Review of B. F. Skinners Verbal Behavior, Language, 35, pp. 26-58.
Davidson, Donald 1967a: Truth and meaning, Synthese, 17, pp. 304-323.
Davidson, Donald 1967b: The logical form of action sentences, in N. Rescher (ed.), The Logic of
Decision and Action. Pittsburgh: University of Pittsburgh Press.
Davidson, Donald 1973-74: On the very idea of a conceptual scheme, Proceedings and Addresses of
the American Philosophical Association, 47, pp. 5-20.

43
Davidson, Donald 1977: The method of truth in metaphysics, in P. A. French, T. E Uehling, Jr., and H.
K. Wettstein (eds.), Midwest Studies in Philosophy, 2: Studies in the Philosophy of Language.
Morris: University of Minnesota Press.
Dennett, Daniel 1981: Brainstorms: Philosophical Essays on Mind and Psychology. Brighton:
Harvester.
Dummett, Michael 1958-9: Truth, Proceedings of the Aristotelian Society, 59, pp. 141-162.
Dummett, Michael 1973: Frege: Philosophy of Language. London: Duckworth.
Dummett, Michael 1975: Wangs paradox, Synthese 30, pp. 301-324.
Dummett, Michael 1978: Can analytical philosophy be systematic, and ought it to be?, in his Truth
and Other Enigmas. London: Duckworth.
Dummett, Michael 1984: An unsuccessful dig, in C. Wright (ed.), Frege: Tradition and Influence.
Oxford: Blackwell.
Dummett, Michael 1991: The Logical Basis of Metaphysics. London: Duckworth.
Evans, Gareth 1982: The Varieties of Reference, ed. J. McDowell. Oxford: Clarendon Press.
Fine, Kit 1975: Vagueness, truth and logic, Synthese 30, pp. 265-300.
Fodor, Jerry 1975: The Language of Thought. New York: Crowell.
Grice, H. P. 1961: The causal theory of perception, Aristotelian Society, sup. 35, pp. 121-52.
Grice, H. P. 1975: Logic and conversation, in D. Davidson and G. Harman (eds.), The Logic of
Grammar. Encino: Dickenson.
Grice, H. P., and Strawson, P. F. 1956: In defence of a dogma, Philosophical Review, 65, pp. 141-158.

44
Hawthorne, John 2004: Knowledge and lotteries. Oxford: Oxford University Press.
Kripke, Saul 1963: Semantical considerations on modal logic, Acta Philosophica Fennica, 16, pp. 8394.
Kripke, Saul 1972: Naming and necessity, in D. Davidson and G. Harman (eds.), Semantics of Natural
Language. Dordrecht: Reidel. Revised version with editions published as his Naming and
Necessity. Oxford: Blackwell, 1980.
Kripke, Saul 1976: Is there a problem about substitutional quantification?, in G. Evans and J.
McDowell (eds.), Truth and Meaning. Oxford: Oxford University Press.
Kripke, Saul 1982: Wittgenstein on Rules and Private Language: An Elementary Exposition. Oxford:
Blackwell.
Kripke, Saul 2013: Reference and Existence: The John Locke Lectures. Oxford: Oxford University Press.
Lewis, David 1968: Counterpart theory and quantified modal logic, Journal of Philosophy, 65, pp.
113-126.
Lewis, David 1969: Convention: A Philosophical Study. Oxford: Blackwell.
Lewis, David 1974: Radical interpretation, Synthese, 23, pp. 331-344.
Lewis, David 1975: Languages and language, in K. Gunderson (ed.), Language, Mind, and
Knowledge. Minneapolis: University of Minnesota Press.
Lewis, David 1983a: Philosophical Papers, Volume I. Oxford: Oxford University Press.
Lewis, David 1983b: New work for a theory of universals, Australasian Journal of Philosophy, 61, pp.
343-377.

45
Lewis, David 1986: On the Plurality of Worlds. Oxford: Blackwell.
Montague, Richard 1970: English as a formal language, in B. Visentini et al. (eds.), Linguaggi nella
Societ e nella Tecnica, Milan: Edizioni di Comunit.
Montague, Richard 1973: The proper treatment of quantification in ordinary English, in J. Hintikka,
J. Moravcsik, and P. Suppes (eds.), Approaches to Natural Language: Proceedings of the 1970
Stanford Workshop on Grammar and Semantics. Dordrecht: Reidel.
Nagel, Thomas 1974: What is it like to be a bat?, Philosophical Review, 83, pp. 435-450.
Peacocke, Christopher 1992: A Study of Concepts. Cambridge, Mass.: MIT Press.
Prior, A. N. 1957: Time and Modality. Oxford: Clarendon Press.
Putnam, Hilary 1978: Meaning and the Moral Sciences. London: Routledge & Kegan Paul.
Putnam, Hilary 1994: Sense, nonsense, and the senses: an inquiry into the powers of the human
mind, Journal of Philosophy, 91, pp. 445-517.
Quine, Willard van Orman 1951: Two dogmas of empiricism, Philosophical Review, 60, pp. 20-43.
Quine, Willard van Orman 1960: Word and Object. Cambridge, Mass.: M.I.T. Press.
Quine, Willard van Orman 1969: Foreword to Lewis 1969.
Rorty, Richard 1979: Philosophy and the Mirror of Nature. Princeton: Princeton University Press.
Salmon, Nathan 1982: Reference and Essence. Oxford: Blackwell.
Searle, John 1969: Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge
University Press.
Stanley, Jason 2005: Knowledge and Practical Interests. Oxford: Clarendon Press.

46
Strawson, P. F. 1950: On referring, Mind, 59, pp. 320-324.
Strawson, P. F. 1952: Introduction to Logical Theory. London: Methuen.
Strawson, P. F. 1959: Individuals: An Essay in Descriptive Metaphysics. London: Methuen.
Strawson, P. F. 1971: Meaning and truth, in his Logico-Linguistic Papers. London: Methuen.
Strawson, P. F. 1977: Analitika filozofija, Theoria (Belgrade), 12, pp. 61-71.
Strawson, P. F. 1992: Analysis and Metaphysics: An Introduction to Philosophy. Oxford: Oxford
University Press.
Strawson, P. F. 1998: Intellectual autobiography, in L. Hahn (ed.), The Philosophy of P.F. Strawson.
Chicago: Open Court.
Wiggins, David 1976: Truth, invention, and the meaning of life, Proceedings of the British Academy,
62, pp. 331-378.
Williamson, Timothy 1988: First-order logics for comparative similarity, Notre Dame Journal for
Formal Logic, 29, pp. 457-481.
Williamson, Timothy 1984: The infinite commitment of finite minds, Canadian Journal of
Philosophy, 14, pp. 235-255.
Williamson, Timothy 1994: Vagueness. London: Routledge.
Williamson, Timothy 2013: Modal Logic as Metaphysics. Oxford: Oxford University Press.

Williamson, Timothy 2014: Replies to Bricker, Divers, and Sullivan on Modal Logic as
Metaphysics, Philosophy and Phenomenological Research, 88, pp. 744-764.
Wilson, John Cook 1926: Statement and Inference. Oxford: Clarendon Press.
Wright, Crispin 1975: On the coherence of vague predicates, Synthese, 30, pp. 325-365.